Measuring Engagement, 2014: model of the lobby

The model: drag to turn, scroll to zoom. It is a measuring tool and not a portrait, so it has no textures or people.

The lobby is the hardest stretch of the walk to place: a large, shiny room, crossed quickly, with few details that stay still for the software that tracks the camera.

What it is built from

• The plan. A ground floor plan published online, read at 2.33 in to its pixel. It is not reproduced here. Seven columns at 342.4 in tie it to the office’s floor plan, since the shafts and steel do not move between floors.
• The hall. Every wall on the building’s grid, from the plan: 161 in between the elevators, three door openings of 42 in on each side, and glass to the lobby. The pool’s curb is 180 by 295 in.
• The dropped panel over the pool. Centred over the pool, as TJ said, with its own size: 205 by 340 in, its underside 168 in up. It comes from two photographs TJ shared, placed by the pool’s known size, which fit to 2.1 pixels. Its length is the softest number here.
• The floor. Silver lines about 57 in apart, read from 35 frames flattened onto the floor at the lobby’s west end, and the floor mats, which are in the 2014 video only. TJ pointed out both.
• The luminous ceiling. 180.5 in up at the west glass, from 9 reads ranging from 178 to 184.5 in. Its grid is centred on the pool, as TJ thought.

• Radiant I (1958), by Richard Lippold, over the pool. 22 rods meeting at a hub 72 in up: six along the axes, eight at 45° in the two upright planes and eight on the diagonals. The tips are triangulated from three photographs published online, to a few inches, and read against a published top view and TJ’s stills. Fans of wire run from each end of the long rod up to the panel and down to the pool. The number of wires in each fan is a count by eye, and the web at the centre is drawn to the plan’s octagon, not wire by wire.
• Icehenge, Frank Gehry’s green glass security desk, added about 2012–13 (sources differ). Its footprint only, about 126 by 215 in, from the frames and one still. The plan draws a smaller desk there.

How the video is placed in it

One size, one rotation and one shift place the lobby’s reconstruction, 388 camera positions from 2:50 to 4:00, using 33 points read by hand in 9 frames: the pool’s corners, a column and the west wall’s mullions. They miss by a median of 5.2 pixels. No frame can be bent to fit, since all share the same numbers.

The way back could not use that fit. Walking the other way, the camera sees the room’s other sides, and its rebuild never joined the way out’s, even at ten frames a second. It also bends: one fit misses its reads by over 200 pixels. So the way back is anchored frame by frame. Seven frames are each solved alone, from crisp corners read by hand (the pool’s curb, the glass at the way in, the mat’s far end and the elevator door’s opening), with the eye held at TJ’s height. Between them the rebuild gives the head, and it gives the place only where its motion is believable, at a walking pace and within 2 ft of the straight line.

Checks the fit was not told

• Eye height. It comes out at 69.0 in, against TJ’s measured 69.25 in. This rests on one estimate: the sill under the glass is taken as 5 in by eye. With no sill, the eye reads 65 in.
• The door. The path passes 23 in from the centre of the door TJ leaves by, at 3:12. The door was not one of the points read.
• The hall. From 2:50 to 2:56 the path keeps within 30 in of the hall’s centre line, a median of 24 in. The hall was not in the fit either.
• Outside. The base of the corner column lands on the real one, and the same numbers carry on to the street corners, with no street landmark in the fit.
• The way back, against what it was not told. The path goes through the revolving door’s drum, round the desk and along the mats, and into the elevator through the door’s opening. The hall’s floor edges land on the real ones in the frames from 18:29 to 18:34.
• The way back’s method, on the way out. Solving the way out’s frames one at a time lands a median of 0.3 ft from its fit, the worst 1.0 ft, with the head within about 1°, though those frames fed the fit too. From only two reads a frame lands a median of 1.1 ft away, but the worst is 7.1 ft: two reads give distance far better than position to the side.

What went wrong along the way

• The way back was not measured at first. For a time, 18:14 to 18:38 was straight lines between six points set by hand, one of them cutting across the desk. TJ noticed. Labelling it the weakest stretch was not enough.
• The elevator bank’s black panels were read as its piers. They are narrower than the plan’s piers, and reads on them gave answers that did not fit. The door openings are what the plan fixes.
• The first plan was not followed closely enough. Walls drawn at an angle to the building’s grid were wrong, as TJ pointed out, and there are none.
• The panel was in the wrong place. Frames looking straight down the hall separate position from height poorly, so the first answer was about 4 ft too far east. TJ’s stills corrected it.
• One floor mat was three. They stand around the desk, on its north, east and west sides, as TJ said.
• The ceiling grid was about 11 and 33 in off when read from frames looking down the hall, until TJ’s stills were used.

Estimated, not measured

• The glass sill, 5 in, by eye. The revolving doors’ height, about 100 in. The elevator doors’ height, about 156 in, and the header over the way in, about 162 in up, each from one photograph. The great columns stop at 25 ft, which is not measured.
• The ceiling’s grid is measured south of the pool and carried across. North of the screen nothing is measured: only the plan’s walls.
• The plan is from about 2015, and the video from 2014. The frames agree with it wherever they are drawn together, but a difference could hide where the video does not look.
• The way back is mostly straight lines between solved frames: about 16 of its 22 s, at 5 to 6 ft/s. Three of the seven frames rest on two reads each, so their position to the side is the softest number: the hall’s floor edges in the frames are the check on it. The frames from 18:20 to 18:22 are too blurred to read.
• Which elevator car TJ rode on the way back is a reading from the frames.

Sculpture: Richard Lippold, Radiant I, 1958. Desk: Frank Gehry, Icehenge. Both are their artists’ work, and are modelled here only as far as placing the video needs. The walk and its instrumentation were a team effort by Iota, at SapientNitro. This model was made in 2026, from a published ground floor plan, TJ’s memory and stills, photographs published online, and the Glass video. None of the images is reproduced here.